National Vital Statistics Reports Volume 68, Number 3 March 21, 2019
Drug Overdose Deaths Involving Fentanyl, 2011–2016 by Merianne Rose Spencer, M.P.H., Margaret Warner, Ph.D., and Brigham A. Bastian, B.S., National Center for Health Statistics; James P. Trinidad, M.P.H., M.S., U.S. Food and Drug Administration; and Holly Hedegaard, M.D., M.S.P.H., National Center for Health Statistics
Abstract reporting over the study period did not change the trend patterns observed. Objectives—Fentanyl, a synthetic opioid, has been Conclusions—This report illustrates the rise in drug increasingly identified in drug overdose deaths. This report overdose deaths involving fentanyl from 2011 through 2016 describes trends in drug overdose deaths involving fentanyl by nationally, and by age, sex, race and ethnicity, and public health demographic characteristics and geographic regions from 2011 region. Understanding national trends and patterns for drug through 2016. overdose deaths involving fentanyl may inform public health Methods—Drug overdose deaths were identified from policies and practices. the National Vital Statistics System—Mortality (NVSS–M) multiple cause-of-death files (2011–2016) using International Keywords: National Vital Statistics System • national trends • Classification of Diseases, 10th Revision underlying causes of mortality • death certificate literal text death (codes X40–X44, X60–X64, X85, or Y10–Y14). NVSS–M records for drug overdose deaths were linked with literal text Introduction from death certificates. Drug overdose deaths involving fentanyl were identified using a methodology established collaboratively Fentanyl is a synthetic opioid that has been involved by the National Center for Health Statistics and U.S. Food and increasingly in drug overdose deaths (1–11). In 2011 and 2012, Drug Administration—referred to as the Drugs Mentioned fentanyl was involved in roughly 1,600 drug overdose deaths with Involvement (DMI) methodology—supplemented with each year, but from 2012 through 2014, the number of drug search terms identified using text analytics software. Fentanyl overdose deaths involving fentanyl more than doubled each year involvement was determined by the presence of any string term (1). State-based analyses suggest that the recent increase in or phrase listing fentanyl, or any fentanyl metabolite, precursor, overdose deaths involving fentanyl may be related to increased analog, or misspelling identified in the death certificate literal text availability of illicitly manufactured fentanyl (3,4,9–11). fields (i.e., the causes of death from Part I, significant conditions Illicitly manufactured fentanyl includes various analogs (e.g., contributing to death from Part II, and a description of how acetylfentanyl, carfentanil, furanyl fentanyl) that vary in potency the injury occurred). Trends were evaluated using the National and can be lethal in very low concentrations (3). Cancer Institute’s Joinpoint Regression Program. National mortality statistics on drug overdose deaths have traditionally used the Results—The number of drug overdose deaths involving International Classification of Diseases, fentanyl was stable in 2011 (1,663) and 2012 (1,615), and began 10th Revision (ICD–10) to classify and monitor drugs involved to increase in 2013, rising to 18,335 deaths in 2016. The age- in deaths (12). However, ICD–10 is limited to broader categories adjusted rate increased from 0.5 per 100,000 standard population (e.g., synthetic opioids other than methadone) that make it in 2011 to 5.9 per 100,000 in 2016, with the increase starting in difficult to identify deaths involving specific drugs of interest 2013 (0.6 in 2013 to 1.3 in 2014 and 2.6 in 2015). Numbers and (e.g., fentanyl or illicitly manufactured fentanyl analogs). rates increased for all sex, age, and racial and ethnic subgroups, The National Center for Health Statistics (NCHS) and the U.S. and most public health regions. Adjustment for improved drug Food and Drug Administration (FDA) collaboratively developed methods, referred to as the Drugs Mentioned with Involvement (DMI) methodology, to analyze literal text data from death
U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES Centers for Disease Control and Prevention National Center for Health Statistics National Vital Statistics System
NCHS reports can be downloaded from: https://www.cdc.gov/nchs/products/index.htm. This report was revised on April 3, 2019, to correct for the inadvertent omission of North Carolina from the list of states in Region 4. 2 National Vital Statistics Reports, Vol. 68, No. 3, March 21, 2019 certificates to identify specific drugs involved in deaths (13). A are assumed to be involved in the death unless contextual recent study using the DMI methodology found that fentanyl was information suggests otherwise (13). For example, “METHICILLIN the ninth most frequently mentioned drug involved in overdose RESISTANT STAPHYLOCOCCUS AUREUS INFECTION” does deaths in 2013 and the most frequently mentioned drug in 2016 not suggest drug involvement in mortality, but rather a type of (2). It also showed that deaths involving fentanyl often involved bacterial infection. Similarly, the phrase “NOT DRUG RELATED” other drugs. For example, in 2016, nearly 70% of deaths clearly indicates that drugs were not involved in the death even involving fentanyl also involved one or more other drugs, such though “DRUG” is included in the phrase. as heroin or cocaine (2). To supplement the DMI methodology, SAS Contextual This report describes the trends in drug overdose deaths Analysis software (18) was used to identify additional search involving fentanyl from 2011 through 2016, using mortality data terms. Linguistic rules were defined to search for new drug from the National Vital Statistics System linked to literal text misspellings; precursors; metabolites; analogs of fentanyl, information from death certificates. Temporal trends, geographic including those containing special characters, numbers, patterns, and differences among demographic groups are and dashes; and to identify forms that may be illicit (e.g., presented. acetylfentanyl, carfentanil) or are brand or prescription forms of fentanyl (e.g., duragesic, sufentanil, remifentanil). The additional Methods search terms identified using the contextual software were incorporated into the DMI methodology and are described in the Technical Notes. With the expanded list of search terms, the total Data sources number of drug overdose deaths identified as involving fentanyl This descriptive study analyzed National Vital Statistics was slightly higher than previously reported (7 deaths in 2011, System mortality (NVSS–M) multiple cause-of-death data from 10 deaths in 2012, 14 deaths in 2013, and 23 deaths in 2014) 2011 through 2016. NVSS–M contains information extracted (1). from death certificates on cause of death and demographic and geographic factors (14). The study population was limited to U.S. Analysis residents. Drug overdose deaths were identified using ICD–10 For this report, the label “drug overdose deaths involving underlying cause-of-death codes X40–X44, X60–X64, X85, and fentanyl” includes drug overdose deaths involving fentanyl, Y10–Y14. These underlying-cause codes identify the deaths due whether prescription or illicitly manufactured, as well as deaths to acute toxicity from drugs (i.e., drug overdose) rather than to involving any fentanyl metabolites, precursors, or analogs as chronic exposure leading to death (e.g., liver toxicity) or adverse identified in the death certificate literal text (see Technical Notes). effects from therapeutic or prophylactic dosages of drugs. Drug The numbers and rates for drug overdoses involving fentanyl overdose deaths include all intents (i.e., unintentional, suicide, were calculated and compared by demographic and geographic homicide, and undetermined intent). Use of these underlying region. Bridged-race vintage postcensal resident population cause-of-death codes is consistent with other NCHS publications estimates were used to calculate death rates (14). Age-adjusted on drug overdose deaths and facilitates comparisons with other death rates were calculated using the direct method and the 2000 analyses using ICD–10-coded data (1,2,15). standard U.S. population (14). Unless otherwise noted, rates in NVSS–M records for drug overdose deaths were linked the text refer to age-adjusted rates. to literal text data from death certificates. The literal text is the Geographic patterns in overdose deaths involving fentanyl written information provided by the medical certifier, usually are presented by the U.S. Department of Health and Human a medical examiner or coroner in the case of drug overdose Services (HHS) 10 public health regions. These regions are used deaths, that describes the cause of death as well as other factors for public health prevention, preparedness, and agency-wide or circumstances that contributed to the death (16,17). Literal coordination of HHS programs and policies (19). These regions, text from three fields of the death certificate—the causes of excluding U.S. territories, are: death from Part I, the other significant conditions contributing to death from Part II, and the description of how the injury • Region 1—Connecticut, Maine, Massachusetts, New occurred—was analyzed to identify the specific drugs involved Hampshire, Rhode Island, Vermont in the overdose death. • Region 2—New Jersey, New York, New York City • Region 3—Delaware, District of Columbia, Maryland, Identifying drug overdose deaths involving Pennsylvania, Virginia, West Virginia fentanyl • Region 4—Alabama, Florida, Georgia, Kentucky, Mississippi, North Carolina, South Carolina, Tennessee This report used the DMI methodology (13) informed by • Region 5—Illinois, Indiana, Michigan, Minnesota, Ohio, the results from an exploratory analysis of the literal text using Wisconsin SAS Contextual Analysis software (18). The DMI methodology • Region 6—Arkansas, Louisiana, New Mexico, Oklahoma, (13) searches the literal text fields of NVSS–M data for mentions Texas of drugs and for terms that provide context about involvement • Region 7—Iowa, Kansas, Missouri, Nebraska of the drug in the death (i.e., whether the drug contributed to the death). Drugs mentioned in the death certificate literal text National Vital Statistics Reports, Vol. 68, No. 3, March 21, 2019 3
• Region 8—Colorado, Montana, North Dakota, South Results Dakota, Utah, Wyoming • Region 9—Arizona, California, Hawaii, Nevada • Region 10—Alaska, Idaho, Oregon, Washington Number and percentage of drug overdose deaths involving fentanyl Trends in rates for drug overdose deaths involving fentanyl were evaluated using the National Cancer Institute’s Joinpoint Annually, the number of drug overdose deaths involving Regression Program Version 4.6.0.0 (20). Joinpoint software fentanyl was stable in 2011 (1,663) and 2012 (1,615), and fitted weighted least-squares regression models to the rates began to increase in 2013 (1,919), rising to 18,335 deaths in on the log transform scale. Allowing one observed time point 2016 (Table 1). From 2013 through 2016, the number of deaths at each end and two for the middle line segments, the Grid approximately doubled each year. From 2011 through the third Search Algorithm searched for a maximum of two joinpoints at quarter of 2013, there were fewer than 500 fentanyl-involved an overall alpha level of p < 0.05 (21). Pairwise comparisons deaths per quarter (Figure 1, Table 1). Beginning in the last of trend segments among subgroups were also analyzed, with quarter of 2013 through 2016, the number of deaths involving Bonferonni adjustment (20). Rates between subgroups were fentanyl increased nearly every quarter to more than 5,800 compared using a z test at the 0.05 level of significance (14). deaths. Unless otherwise stated, any mention of an increase in rates indicates a statistically significant change. Crude and age-adjusted rates of drug overdose deaths involving fentanyl Assessment for improved reporting on death The age-adjusted rate of drug overdose deaths involving certificates fentanyl was 0.5 per 100,000 in 2011 and 2012, and approximately The ICD–10 multiple-cause codes T36–T50.8 provide doubled each year from 0.6 (2013) to 1.3 (2014) to 2.6 (2015) information on the types of drugs or drug classes involved to 5.9 (2016) (Figure 2, Table 2). From 2011 through 2013, there in the death. The ICD–10 multiple-cause code T50.9 was no statistical change in the age-adjusted rates, but from indicates a nonspecific reference to a drug (e.g., “multidrug,” 2013 through 2016, the rates increased on average by about “polypharmacy,” “drug”). The percentage of deaths with an 113% per year. The crude rates were similar to the age-adjusted underlying cause of X40–X44, X60–X64, X85, or Y10–Y14 that rates for the study period. had a multiple-cause code of T36–T50.8 is a measure of the specificity of reporting drugs or drug classes in drug overdose Rates of drug overdose deaths involving deaths. This measure was used to assess possible changes in fentanyl by sex of decedent reporting of specific drugs and drug classes through the years of the study. The percentage of drug overdose deaths coded with For males, age-adjusted rates for drug overdose deaths were T36–T50.8 increased each year (75.0% in 2011, 76.0% in 2012, stable from 2011 through 2013, and then increased by 125.6% 77.9% in 2013, 80.7% in 2014, 83.1% in 2015, and 85.4% in per year from 2013 through 2016 (Figure 3, Tables 2 and 3). 2016) (21,22). For females, the age-adjusted rates increased exponentially from To assess the possible impact of lower reporting of specific 2011 through 2016. In 2011, 2012, and 2013, the rates for males drugs in the years prior to 2016, an adjustment factor accounting and females were similar at 0.6 to 0.7 per 100,000 for males, for improved reporting of specific drugs over the time period and 0.4 to 0.5 for females. Starting in 2013, the rates diverged, was applied to each age-adjusted rate for drug overdose deaths with the rate for males increasing more rapidly than the rate for involving fentanyl (i.e., by year, sex, age, race and ethnicity, and females. By 2016, the rate for males (8.6) was 2.8 times the rate public health region). The adjustment factor assumed that the for females (3.1). specificity of drug reporting had remained constant from 2011 through 2016 at the 2016 drug specificity rate; the adjustment Rates of drug overdose deaths involving factor was recalculated and applied to each age-adjusted rate fentanyl by age group for the stratified analyses (i.e., demographic and geographic groups). Results from this analysis show similar trends for Exponential increases in rates from 2011 through 2016 both the observed and adjusted values (see Technical Notes). were observed among all age groups (Figure 4, Tables 2 and 3). The following results reflect the numbers and rates based on The largest average annual percent change from 2011 through observed values. The number of drug overdose deaths involving 2016 occurred among young adults aged 25–34 (100.0% per any particular drug in any particular year should be considered year) and 15–24 (93.9% per year). The smallest average annual the minimum number, as there may be additional deaths in percent change occurred among adults aged 65 and over (41.6% which the drug was involved but not specified on the death per year). The rate for adults aged 35–44 was stable from 2011 certificate literal text. All results should be interpreted in light of through 2013, then increased by 123.7% per year from 2013 the improved reporting of specific drugs in the literal text over through 2016. In 2016, the rates were highest among adults aged time. 25–34 and 35–44 at 13.4 and 11.4 per 100,000, respectively. 4 National Vital Statistics Reports, Vol. 68, No. 3, March 21, 2019